GUIDED FILTER BASED IMAGE FUSION ALGO - RITHM IN LIFTING WAVELET DOMAIN

Abstract

This work presents fusion of medical images using guided filter (GF) in lifting wavelet transform domain. The medical images used are CT and MRI. The lifting wavelet transform have been implemented on CT and MRI images to obtain the sub-images i.e. LL, LH, HL and HH subimages. By comparing the LL sub-images of input images, the weight maps have been obtained. Then, the refined weight map has been obtained using the Gaussian filter. Now for obtaining the guided image the Canny edge detector is used. Finally, using the weight maps as the input image of the guided filter and the edge detected image using canny operator as the guided image, guided filter has been designed. The approximation and wavelet coefficients of CT and MRI images are fused according to the weighted fusion rule using refined weight maps. A fused image of CT and MRI has been obtained by the inverse lifting wavelet transform. The proposed method has been compared with Choose-max method, intuitionistic fuzzy inference method and guided filter based wavelet transform method for fusion. The image fusion has been evaluated using entropy, correlation, average gradient, edge strength. Simulation results present the better performance of the lifting wavelet domain based fusion method in terms of these parameters. Also the use of the lifting wavelet transform has the advantage of smaller complexity as compared to the first generation wavelets.